[JohnBee]

Talk is cheap! everyone is forgetting one thing...

Hello jamesm007,

I'd like to cut-in for a moment and offer assurances on the fact that the claim where 'talk is cheap' was not intended as an attack on you personally. What was meant by this specifically, was where theories(or explanations) of this nature are unlikely to prove conclusive with regards to answers. Which isn't to say that it doesn't help establish credibility toward a specific claim. However... in cases such as these(assertions), the only way to address the issue, is to devise methods from which to obtain tangible conclusions.

So for what its worth, I hope that you didn't misinterpret my reference as an attack on you or what you posted. Because, had you not written this, I don't think I would of bothered to write this thread. But, in fact, it was the logical presentation provided in your comment that motivated me to post a request on the matter.

On the issue of talk vs testing...
I've come to the conclusion that the only way to deal with claims such as these, is by addressing them head-on.

This is risky business imo. and(as can be observed)... not always understood or appreciated by certain people. However... in defense of what is right, I'd like to remind everyone that trolls 'are neither interested nor committed to the truth'. Which is precisely what threads of this nature are intended to do.

For what its worth, I think the comment and logic offered by yourself makes good sense. However... by leaving it as such(a standing theory) then it will likely never get addressed or substantiated as anything but a theory. Which is precisely why and how such rumors just seem to linger-on forever.

-

Having said all that... I also want to add that I appreciate your objective opinions and efforts on the issue. And without sounding like an opportunist, I was wondering if you might have a K-5 and another body that could be used to produce some simple contrasting samples with? I know there is always the potential of observing the phenomenon of PF/CA changes in some of the public testing samples. However, given the potential for the infinite variances that can occur in remote testing, I'm not sure how reliable the results can be.

Based on this, I'm thinking that the path of less resistance toward a conclusion might be achieved by conducting tests with the same lens across different bodies in the same setting.
What do you think?

And this comment of course is not limited to jamesm007 alone.
I invite anyone with an interest in exploring the issue to chime-in and offer their opinion and participation toward the issue.

---

Last edited by JohnBee; 11-29-2010 at 07:20 PM.

[jamesm007]

Thanks John ...

I do know I kinda (kinda...) wrote some bold claims and just left it stand!!! To be honest I was hoping GordonBGood or Falk would jump in and set things straight or at least say, "thats not possible". To my surprise no one said anything and I could not understand why? After I did more research and can't come to any conclusions but one. For some reason the K-5 shows PF/CA higher than I am use to seeing, maybe! It could very well be because of its high resolving power and focus is critical with this camera. In fact this I believe as much as anything else. Or lets just say "I don't know"


I am the type who wants to know. Kinda why I did what I did. Believe me I am pure science in my thinking. I know I don't know. In this case I don't know why or even if the K-5 has higher than normal CA/PF. I don't even know if my theory's are possible in this case. This one is tough... So I guess my putting that out there was ...well... wrong.

I do read your posts all the time and find them interesting and was very happy with how far you took the K20D (high ISOs) nice . And do appreciate and support your efforts (which are strong) to keep trolls from walking all over Pentax fans 'fun'.

I do think tests although they will be informal will let us know if we should pursue this further or not. Example, if the K-5 tends to show more CA/PF in back to back tests with a K-7. Well we will have to dig for an answer.

Anyways appreciate you writing, thanks.

BTW no K-5 for me till around Jan-Feb...

[Manel Brand]

This PF showing up in just about every K-5 shoot is starting to annoy me, I was all set to get the K-5 but for the first time I'm beginning to hesitate.
Is PF the k-5s Achilles heel?

Don't judge K-5 performance from what you see posted on the web. So far, I've seen horrible images and I don't believe that can be K-5 fault. The same horrible images can be seen captured from any other camera on the market. Some dubious people here and elsewhere fancy themselves as photographers and testers but they don't know how to take advantage of the gear they have. Better for them to go back to that crappy P&S and stop posting Bs. On the other hand, I've seen some great shots by other members taken by the very same camera and that just reassures my positive expectations regarding K-5's performance on paper.

[Class A]

When I'm looking at a pic, what's the difference between pf and ca?

There is none.

Purple fringing is caused by dispersion. So PF is first and foremost a lens issue.

Many Pentax lenses are known to have quite a bit PF so maybe that's the reason why one sees more K-5 shots with PF than one sees D7000 shots with PF.

The only way I could image that the K-5 would be worse than say the K-7 regarding showing PF is a) it is a better tool and shows lens imperfections in a more pronounced manner or b) it is more sensitive in the "(ultra)violet and (infra)red regimes, where lenses tend to be poorly corrected for chromatic aberration."

In any event, whatever the outcome of the test, it won't change my mind that I'll get the best Pentax DSLR ever. I didn't go for the K-7 because of the Samsung sensor but there is nothing about the K-5 that says "stay away".

BTW, I don't know what the guys at IR did. They published horrible K-5 images without noticing that something was wrong with them. Their images that were allegedly taken with a Sigma 70/2.8 showed PF. Maybe the images were from a different lens, maybe they dropped the lens on concrete a couple of times. The Sigma 70/2.8 is an excellent lens whose "Lateral CAs (color shadows at the image borders) are very well controlled within the full format testing scope. There're only very minor traces of purple fringing at f/2.8.". The IR images looked nothing like I would expect from this lens (which I own).

I like the IR reviews a lot. The K100D review from IR convinced me to go for the K100D (among other things). I still think highly of IR but I have no idea what went wrong with their K-5 shots, but something went seriously wrong.

---

Last edited by Class A; 12-03-2010 at 08:26 PM.

[jamesm007]

Class A wrote:
BTW, I don't know what the guys at IR did. They published horrible K-5 images without noticing that something was wrong with them.



At least they IR used a good lens this time. Many don't know in testing the K20D IR used a bad copy (very soft on right side) of the Sigma lens. This allowed trolls and others to prove their case in attacks by just showing the comparo meter as I will show below.

This is showing the round scale and brush. Two pics, the one on the left is much softer than the right. The left is the K20D the right the Nikon D300. The iso I used was 100 for the Pentax and 200 for the Nikon as I believe thats the Nikon D300 native ISO. In any case the results are the same and against any camera. The K20D will lose in any test that used the Sigma lens. In fact any Pentax they tested up until the K-7. How many people use this meter to judge? The truth is the K20D and D300 have near the same resolution; not this glaring difference!

Here is a quote from IR after being told they must have a bad lens. They agreed to look and found they did have a bad Pentax version of the Sigma lens. Really it affects the whole picture. But for years IR used a bad copy of the Sigma lens on Pentax dSLRs.

"All cameras were equipped with our Sigma 70mm f/2.8 EX DG Macro reference lenses, stopped down to f/8 for maximum sharpness. (We've replaced our Pentax mount version with one that doesn't have the softness our previous copy had in the right-hand side. That's why the scale crop from the K20D is a bit soft compared to the others, as it used the older lens."

Pentax K-7 Digital Camera Raw - Full Review - The Imaging Resource!
(on the bottom of the page)

A bit soft!!!



Back in the K10D and K20D day testers did not even bother to tweak the settings. The K10D was too soft, the K20D to bright and too much noise in the upper ISOs in JPEG (Testers never showed its RAW except with a bad lens). No one knew how good these cameras were except us owners as John who showed well how good they are

So its not too surprising IR did not notice how bad the K-5 pics were? Just a bit of history.

[Class A]

So its not too surprising IR did not notice how bad the K-5 pics were? Just a bit of history.

Thanks for that!
Bit of a sad story, eh?

Perhaps I really only like the talents of Shawn Barnett, displayed in his K100D review and the measurement department of IR isn't that great? At least IR often praised Pentax cameras and didn't slam them for stupid things like DPR did.

[gazonk]

The issue of the K-5 being especially prone to PF popped up again today as the Phoblographer published the third part of his K-5 review and complained about it:
Field Review: The Pentax K-5 (Day 3) The Phoblographer

[JohnBee]

The issue of the K-5 being especially prone to PF popped up again today as the Phoblographer published the third part of his K-5 review and complained about it:
Field Review: The Pentax K-5 (Day 3) The Phoblographer

Wow those samples seem alarming!

Though without knowing the settings(aperture etc) and some toe to toe comparisons(same lens, different camera), I honestly don't know how significant the samples can be.

Having said that, I am still hoping someone will take the initiative and do a comparisson with the K-5 against other camera's to show if there is a change(if any). Which I think would be the best way to show that.

though the PF in those samples are just short of horrific!
If it were me, I would try the lens on another body just to be sure, but it looks like it needs to go back imo.
There's no reason for a * lens to perform this badly at any aperture.

[panoguy]

Wait, so Longitudinal Chromatic Aberration is thought to be worse on a particular sensor?

I understand you can't test this yourself, either JohnBee or jamesm007, because neither of you have the camera in question, but obviously there are now K-5 samples out there that show this lens aberration.

So are we waiting for someone with a K-5 (and say a K-7, to vary the sensor) to test a single lens (that has lots of LCA fringing, say a 77mm Limited) with a consistent high-contrast scene and at various apertures (since LCA varies with aperture) and process the raw files identically to see which sensor shows this lens-based aberration to a greater degree?

While we're at it, can we get a Nikon D7000 thrown in to see if its the sensor or the filters that each company uses? Oh wait, the lens won't mount... and using two different lenses voids the quantifiability of the test.

Interesting, but not alarming. Good luck!

[snostorm]

FWIW, a long time ago, I noticed that the K10 was considerably more prone to PF than the K20. I once shot a back to back test using my FA* 300/4.5 wide open at a branch of a pine tree silhouetted against a bright sky and this showed pretty conclusively that the sensor used made the difference. Can't find the pics now. . . Just tried a similar test with the K10, K-7, and K-5, but it's very overcast and I couldn't get any of them to show the effect.

BTW, this lens is not particularly prone to showing this, but it occurs.

I've found that CA/PF are both lens and sensor dependent as my Sigma EX 180/3.5 DG Macro is the best at suppressing both. . .

I'm primarily a birder, so I run into situations where this occurs on a regular basis. I've found an effective way for me to deal with it in PP, so this is not a high priority for me when choosing to pruchase a body.

Scott

[Dan]

FWIW, a long time ago, I noticed that the K10 was considerably more prone to PF than the K20. ...

I found exactly the same thing. In shooting the same scene with the same lens (DA 21) and same settings, the K10D exhibited much worse PF than the K20D. The difference was stark. Indeed, I found that all of my lenses that were prone to PF with the K10D performed much better in this regard with the K20D. This is an underappreciated characteristic of the K20D. Based on my experience, I do not doubt that the sensor assembly can make a difference when it comes to PF.

. . . Just tried a similar test with the K10, K-7, and K-5, but it's very overcast and I couldn't get any of them to show the effect. ...

I hope that you are able to run a test under challenging conditions, using a lens that is prone to PF. I am very interested in your results. At this point it seems premature to reach any conclusions, good or bad.

Dan

[Class A]

The issue of the K-5 being especially prone to PF popped up again today as the Phoblographer published the third part of his K-5 review and complained about it:

What if it was just the lens?

Can we please stop the "the K-5 produces PF" news, unless someone can demonstrate that indeed the sensor plays a part (e.g., by comparing to a K-7)?

[gazonk]

What if it was just the lens?

The DA*55? One shouldn't think so.

Can we please stop the "the K-5 produces PF" news, unless someone can demonstrate that indeed the sensor plays a part (e.g., by comparing to a K-7)?

Please don't shoot the messenger, I'm just citing a review. It would indeed be good to compare with a different camera, but with the same lens.

(It's possible that different sensors react differently. I don't think it's pixel bleeding we see here, but isn't there a theoretical possibility that the sensors react differently e.g. to spectral purple light? Or even UV?)

[Class A]

The DA*55? One shouldn't think so.

Oh yes, it shows purple fringes. Blende8 didn't test for it specifically but you can easily see (traces of) it in his test shots.

Also check out this "magenta feast" by the DA* 55 as featured in the SLR Lens & Camera review of the DA* 55.

Please don't shoot the messenger, I'm just citing a review.

Messengers forwarding true messages must not be shot. Messengers who make up news or help to purport unverified rumours deserve to be shot. Sorry.

[falconeye]

Wow those samples seem alarming!

Though without knowing the settings(aperture etc) and some toe to toe comparisons(same lens, different camera), I honestly don't know how significant the samples can be.

Having said that, I am still hoping someone will take the initiative and do a comparisson with the K-5 against other camera's to show if there is a change(if any). Which I think would be the best way to show that.

though the PF in those samples are just short of horrific!
If it were me, I would try the lens on another body just to be sure, but it looks like it needs to go back imo.
There's no reason for a * lens to perform this badly at any aperture.

JohnBee, the first crop clearly is Bokeh CA from the lens. You can see this by going to the full image in the top of the article and look at the cropped region. it is in the foreground and purple. But there is a second almost identical region in the background (second lamp) which is green.

The son is less clear. It shouldn't be this purple and not if his face is in focus. Maybe, this was indeed from the defect lens with Long.CA.

If Bokeh CA is more visible with the K-5 is unknown ATM. The higher dynamic range - when exploited by shadow compensation - will certainly amplify the visibility of any kind of CA near contrast borders. That's optics.